space-alc.rb

# -*- coding: utf-8 -*-
#!/usr/bin/env ruby

# GET english examples from SpaceAlc

require 'uri'
require 'open-uri'
require 'rubygems'
require 'nokogiri'

class SpaceAlc
  def self.base_uri
    URI 'http://eow.alc.co.jp/'
  end

  def initialize(word)
    @uri = self.class.base_uri
    @uri.path = '/' + word.gsub(/\s+/, '+') + '/UTF-8/'
  end

  def to_hash
    if defined?(@_to_hash)
      @_to_hash
    else
      doc = Nokogiri::HTML(open(@uri, 'User-Agent' => 'Mozilla/5.0'))
      doc.xpath('//div[@id="resultList"]//li').inject({ }) do |@_to_hash, item|
        @_to_hash.merge({ item.xpath('span[@class="midashi"]').inner_text =>
                          item.xpath('div[1]').inner_text })

      end
    end
  end

  def to_portable
    Hash[*to_hash.collect { |word, meaning|
           [word, to_trim(meaning)] if portable?(word)
         }.flatten.compact]
  end

  private

  def portable?(word)
    len = word.split(/\s+/).length
    len == 1 or (len > 3 and len < 6) or (len > 10 and len < 15)
  end

  def to_trim(meaning)
    meaning.gsub /【.*?】|〔.*?〕|｛.*?｝|《.*?》|◆.*$/, ''
  end
end